With the deepening of physical education reform, students’ physical fitness training faces problems such as non-standardized movements and single training method, and the use of data analysis and model construction can effectively improve the effect of physical fitness training and help improve the quality of physical education teaching in colleges and universities. In this study, principal component analysis is used to reduce the dimensionality of students’ physical fitness data, and a training action detection model is constructed by combining time evolution modeling and multi-case learning to standardize students’ physical fitness training actions. Sixty sophomore students from a university were divided into experimental and control groups, and the experimental group applied the action detection model designed in this paper to carry out physical fitness training. The results showed that: the detection accuracy of the three physical training movements (pull-up, 50-meter run, sit-up) were kept above 90%; after the experiment, the indicators of lung capacity (4383±661.29 ml) and sit-up (45±9.07) of the experimental group were significantly better than those of the control group (3317±1307.35 ml, 36±14.28); through the principal component comprehensive evaluation modeling, the comprehensive physical fitness qualities of the experimental and control groups were 2.37 and 1.75, respectively, with significant differences. The study confirms that the physical training movement detection model based on multivariate statistical analysis can effectively identify and correct students’ training movements, improve the accuracy of training, promote the overall improvement of students’ physical fitness level, and provide new ideas for the reform of physical education teaching in colleges and universities.
